1B Parameter LLMs — Page 101

merge_linear_cos0.9fmt0.1_MRL4096_ROLLOUT4_LR1e-6

dpo-llama3.2-gspo-original-200

SmolLM3-DPO-Second-Round-no-think

merge_accfmt_MRL4096_ROLLOUT4_LR5e-7_w0.9_linear

merge_accfmt_MRL4096_ROLLOUT4_LR5e-7_w0.3_linear

merge_cosfmt_MRL4096_ROLLOUT4_LR5e-7_w0.3_linear

merge_lenfmt_MRL4096_ROLLOUT4_LR5e-7_w0.9_linear

merge_lenfmt_MRL4096_ROLLOUT4_LR5e-7_w0.7_linear

merge_lenfmt_MRL4096_ROLLOUT4_LR5e-7_w0.1_linear

merge_accfmt_MRL4096_ROLLOUT4_LR2e-6_w0.9_linear

merge_accfmt_MRL4096_ROLLOUT4_LR2e-6_w0.7_linear

merge_cosfmt_MRL4096_ROLLOUT4_LR2e-6_w0.5_linear

merge_cosfmt_MRL4096_ROLLOUT4_LR2e-6_w0.1_linear

merge_lenfmt_MRL4096_ROLLOUT4_LR2e-6_w0.9_linear

merge_lenfmt_MRL4096_ROLLOUT4_LR2e-6_w0.3_linear

SB_DS1.5B_alpha_1

gemma-3-1b-pt-MED

Qwen2.5-1.5B-Open-R1-GRPO-Crosswords-v7

expert_acc_MRL4096_ROLLOUT4_LR1e-6_step50

Qwen2.5-Math-1.5B-CVAPO-ADAPTIVE-G8

merge_accfmt_MRL4096_ROLLOUT4_LR1e-6_w0.5_pcb

trainorder

math_len_1.5B

pdalma_ctx4_dm1_ce01_pr1_ptll32-1b_s2_ckpt_9_of_10_it311

pdcd200_cptq15_ce01_pr0_ptq25-15b_omi_c100k_200tok_s8_ckpt_9_of_10_it1135

sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_3_of_5

Random_CTPT_final_model

tinyllama-1.1B-sparse-20

Qwen2.5-1.5B-random-weights

jennifer-gemma-3-1b-it

Qwen2.5-1.5B-GRPO-evo-2

llama-sft-proj

chess-qwen

super-chess-qwen

hh_qwen1.5_drpo_gated_fixed_beta

chess-qwen-lora-v1

extractor_abreviaciones

OpenRS-GRPO-S-2

M3PO-baseline-trial4

A2-Model-SFT-RESTA

A2-Model-SFT-DARE-RESTA

asgn2-dare-optimal